Google to Test Machine Learning-Based Feature to Detect Underage Users on YouTube
Google is gearing up to start testing a machine learning-based feature this year that uses computer vision to identify underage users trying to access adult content on YouTube. The "machine learning-based age estimation model" will attempt to determine whether a user is under 18, and if so, apply age-restricted settings to their account.
Announcement Amid Calls for Enhanced Child Safety Measures
The announcement comes as the US Senate considers a bill that would ban pre-teens from using social media. Google’s decision to test this feature highlights the company’s growing concern for child safety and its efforts to provide a more age-appropriate experience for its users.
How the Feature Will Work
According to YouTube CEO Neal Mohan, the ML age restriction feature will use various factors to make its predictions, including the types of videos a user watches, the categories of content they search for, and the age of their account. For instance, if a user frequently searches for topics related to finance or healthcare, it may indicate that they are over 18. Similarly, if an account is over 20 years old, it may also be considered an adult.
Assisting in Age-Appropriate Experience
If the ML model determines that a user is underage, YouTube will apply its standard age restrictions to deliver a more age-appropriate experience. This includes blocking explicit and age-restricted videos from under-18 users. Additionally, Google’s SafeSearch Filter will also be applied to block explicit content from search results.
Implementation and Future Developments
The feature is expected to be tested by the end of this year, with plans to roll it out globally in 2026. This is not the only project Google is working on to enhance child safety. The company is also launching the "School Time" feature on Android phones and tablets, which will allow parents to set time limits and approve contact individuals for their children’s devices. Furthermore, Google’s Family Link app will introduce a new feature that allows parents to approve or deny contacts to add to their children’s devices. This feature is expected to roll out in the coming months.
Competition and Other Efforts
Google is not alone in its efforts to safeguard children online. Meta has also announced plans to use an "adult classifier" tool to identify underage Instagram users posing as adults. This feature will be used to detect and prevent minors from accessing adult content on the platform.
Extended Family Link Features
In addition to the machine learning-based age estimation model, Google is introducing several other features to protect children online. These include:
- School Time: previously limited to smartwatches, will now be available on Android phones and tablets.
- Parents will be able to approve or deny contacts to add to their children’s devices on the Android Family Link app.
- In the spring, parents will be able to add tap-to-pay functionality to their children’s Android phones, allowing them to approve payment card information, add or remove cards, and view transactions.
These efforts demonstrate Google’s commitment to protecting children online and providing a safe and age-appropriate experience for its users.
Source Link